Prostate Cancer Stem Cells A Target for New Therapies

N. J. Maitland(K), S. D. Bryce, M. J. Stower, A. T. Collins

Department of Biology, YCR Cancer Research Unit, University of York, Y010 5YW York, UK email: [email protected]

1 Introduction 156

1.1 Prostate Cancer Therapy: A Historical View 156

2 Transgenic Mouse Models of Prostate Cancer 157

3 Prostate Cancer: A Disease of Epithelial Differentiation 157

4 Stem cell Expansion In Vitro 158

5 Definition of the Stem Cell Phenotype in Prostate 159

6 The Origins of Prostate Cancer 161

7 Isolation of Prostate Cancer Stem Cells 163

8 Gene Expression in Prostate Cancer Stem Cells 167

9 Implications for Prostate Cancer Therapy 170

9.1 Surgical Therapy 172

9.2 Radiotherapy 172

9.3 Medical Therapy 172

9.4 Gene and Immunotherapy 173

References 175

Abstract. Prostate cancer is now a common disease in men over 50 years of age. Medical therapies for prostate cancer are based on discoveries from the mid-twentieth century, and in the long term are rarely curative. Most treatments are directed towards an androgen receptor-expressing, highly proliferative target cell, which does indeed form the vast majority of cells in a prostate tumour. However, by invoking the existence of a cancer stem cell which, like normal epithelial stem cells in the prostate, does not express androgen receptor and is relatively quiescent, the observed resistance to most medical therapies can be explained. The phenotype of the prostate cancer stem cells is that of a basal cell and cultures derived from cancers, but not benign tissues, express a range of prostate cancer-associated RNAs. Furthermore, stem cells purified on the basis of high integrin and CD133 cell surface antigen expression, from an established culture of Gleason 4 (2+2) prostate cancer (P4E6), were able to form multiple intraprostatic tumours in nude mice when grafted orthotopically in a matrigel plug containing human prostatic stroma. The final tumours re-expressed androgen receptor and displayed a histology similar to that of a Glea-son 4 cancer.

1 Introduction

1.1 Prostate Cancer Therapy: A Historical View

Despite intensive study, we still know little about the aetiology of prostate cancer. The patient risk group is well defined, namely elderly men, and as a result most therapies and aetiological studies have centred on male sex hormones. Initially, as a result of the pioneering work of Charles Huggins almost 70 years ago (1941), removal of androgen supply by either orchiectomy or the chemical blockade of adrenal androgen and subsequent direct inhibition of androgen receptor binding to the activated (dihydrotestosterone) form of androgen (since the 1980s) provided almost instant relief from symptoms and longer-term tumour regression (Furr 1996). While these treatments remain palliative, prostate cancer eventually derives resistance to androgen-based therapies and returns in a frequently more aggressive, androgen independent form, which is fatal within 2 years of recurrence on average.

Even the more sophisticated medical therapies, such as optimised taxotere regimes (Tannock et al. 2004; Mackler and Pienta 2006), can do little to stem the inevitability of disease outcome, and the more common anti-mitotic drugs are equally ineffective. This poses a considerable challenge, not only to clinicians, but also to basic scientists. Recently, a first stage in describing the androgen-insensitive phenotype was described by Chen et al. and Holzbeierlein et al. (Chen et al. 2004; Holzbeierlein et al. 2004). Most strikingly, rather than an inactivating androgen receptor (AR) gene mutation a majority of recurrent prostate cancers show amplification of the AR gene, at its locus on the X chromosome (Xq11.2-12), which is present in only one copy in men. In deed, when AR mutations are observed, they frequently result in either enhanced androgen binding or a relaxation of the steroid specificity of the receptor, enabling it to activated gene expression after binding to other steroids, estrogens and even the commoner antagonists used for therapy (Culig et al. 1993; Linja and Visakorpi. 2004). Zero function mutations, such as those found in androgen insensitivity patients, are frequently toxic when re-expressed in prostate epithelial cells (Birnie and Maitland, unpublished data). However, the mechanism whereby the AR gene amplification arises remains unknown.

2 Transgenic Mouse Models of Prostate Cancer

A further role for androgens was provided by transgenic mouse experiments, such as the TRAMP system (Greenberg et al. 1995), where a potent viral oncogene (T antigen from SV40) was linked to a strong promoter (probasin) active in prostate luminal cells in rodents and introduced into the germ line of mice. The resulting TRAMP mice developed premalignant disease which eventually gave rise to carcinomas, which were androgen sensitive. A full description of this system and more recent developments using conditional and total knockouts of known prostate tumour suppressor genes such as PTEN can be found in Winter et al. (2003) and Kasper (2005).

3 Prostate Cancer: A Disease of Epithelial Differentiation

Our starting point was to treat prostate cancers not as an endocrinologi-cal disorder but as an epithelial disease, which had the added complexity of androgen responsiveness, in a proportion of cells. Inherent in most epithelial systems is a pre-programmed differentiation series in which the terminally differentiated cell, which produces the secretory proteins characteristic of the tissue, has resulted from a maturation through one or more epithelial cell intermediates. A major challenge in epithelial biology has been to define the phenotype of these intermediate cells, as they are frequently described by their expression of antigens and proteins, which have little to do with the whole differentiation process. As such, the diversity of cytokeratin expression can provide insights into cellular diversity, and they have been exploited in prostate to establish rudimentary cell lineages (Hudson et al. 2001). The proposed maturation of epithelial cells in the normal human prostate is shown in Fig. 1. From the quoted references it is clear that the two major epithelial compartments in prostatic epithelium, namely the basal and luminal cells, can be subdivided into for example a basal-like transit amplifying cell and perhaps an immature luminal cell. Both of these cell types can be considered to be lineage committed with the resultant luminal cells as the terminal form, lacking both proliferative and long-term survival capacity. Within the true basal compartment, it is now accepted that there exist stem cells with a capacity for unlimited or prolonged self-renewal and the ability to produce at least one type of highly differentiated descendant. It is also accepted that between the stem cell and its terminally differentiated progeny there is an intermediate population of committed progenitors with limited proliferative capacity and restricted differentiation potential, sometimes termed transit amplifying (TA) cells. Furthermore, these normal tissue stem cells (nSC) have special protective mechanisms to resist phenotypic and genetic change such as drug efflux proteins and retention of the parental DNA strand during DNA replication and mitosis. The stability of the nSC may also partly be due to the immediate environment, sometimes called the 'stem cell niche', where the stem cell can sense and react to its contact with basement membrane as well as its spatial contact laterally with other basal cells and apically with more differentiated progeny. Definition of the essential signals for maintenance of 'sternness' has proved an elusive goal, but as methods for separation purification and tissue reconstruction from tissue stem cells have improved, it is becoming clear that these signals are shared with signalling pathways previously recognised to play a role in onco-genesis and indeed normal embryonic development (Pardal et al. 2003; Beachy et al. 2004).

4 Stem cell Expansion In Vitro

The challenge in all studies of this nature has been to achieve purity in the putative stem cell population and the paradox that a stem cell

Fig. 1. Hypothetical epithelial cell differentiation in normal and malignant human prostate. The scheme shows major intermediates in epithelial cell differentiation and indicates possible points at which mutagenesis and hence carcinogenesis can occur. The influence of androgens on the epithelial cells is solely on the most differentiated (luminal) cells, although some prostatic stromal cells also express AR and can influence epithelial cell growth and differentiation

Fig. 1. Hypothetical epithelial cell differentiation in normal and malignant human prostate. The scheme shows major intermediates in epithelial cell differentiation and indicates possible points at which mutagenesis and hence carcinogenesis can occur. The influence of androgens on the epithelial cells is solely on the most differentiated (luminal) cells, although some prostatic stromal cells also express AR and can influence epithelial cell growth and differentiation may sacrifice its 'sternness' when induced to proliferate. Development of stem cell media, and the discovery of small molecules (such as LIF, EGF and FGF2) which can promote expansion of the stem compartment relative to more differentiated forms, has also been a major step forward. However, the complex nature of epithelial tissues has made them more resistant to this type of analysis, relative for example to the haematopoietic system, where markers and lineage determination are more advanced, although as shown by some recent studies in T cell subsets (reviewed in Lefrancois and Marzo 2006) are not terminally defined.

5 Definition of the Stem Cell Phenotype in Prostate

In prostate epithelium, the existence of stem cells was originally proposed by Isaacs and Coffey (1989) based on the available tools at that time. Further refinement of the putative stem cell compartment was possible with the use of defined cytokeratin profiles (De Marzo et al. 1998) However, these markers are of little use for tissue fractionation. What was needed were robust plasma membrane expressed proteins with the ability to define the stem cell compartments. Whilst it was clear that the multiply spliced CD44 antigen had discriminatory power for basal cells over the more abundant and protein content-rich luminal cells, a major breakthrough came from the study of cell adhesion molecules. As first reported by Collins et al. (2001), a heterogeneity can be observed in the basal epithelial layer of normal prostate with respect to the collagen binding (and hence basement membrane associating) a2p1-integrin complex. It was proposed that these integrin 'bright' cells would adhere more rapidly to collagen matrix after tissue dissociation and would result in a cell preparation with enhanced clonogenicity and self-renewal capacity in vivo and in vitro. This was indeed the case, although the resultant preparation was by no means homogeneous. Further enrichment of this small fraction (1%-5%) of the basal cells was achieved by employing the CD133 (originally AC133) antigen (Yin et al. 1997; Corbeil et al. 2000) expression, which was found in a subset of the integrin-bright basal cells. The resultant population exhibited enhanced clono-genicity in vitro and, upon engraftment as a 'micro-prostate' in a matrigel plug containing prostate stromal cells into a subcutaneous site in nude male mice in the presence of androgen, a capacity to produce vestigial prostate glands. These luminal structures now expressed markers of differentiated prostate such as androgen receptor and prostatic acid phosphatase (Richardson et al. 2004).

The requirement for prostatic stromal cells, which also contain an androgen-responsive fraction was absolute, and recapitulated experiments carried out 30 years earlier, when Cunha (1975) showed that co-engraftment of both ovarian and prostatic murine epithelium with male urogenital sinus mesenchyme resulted in development of prostatic epithelium independently of the source of the epithelium (male or female). Indeed, our own in vitro prostate reconstructions confirmed the requirement for prostatic stroma in order to recapitulate the entire differentiation program, including polarisation and secretion of 'prostate-specific' products (Lang et al. 2001).

The a2p1Hi/CD133+/CD44+ population therefore enabled a profound enrichment of the epithelial stem cell population from normal human prostate. However, it constituted a very small fraction of cells from any non-malignant human prostate tissue, and had a limited life span in vitro, despite growth medium optimisation. The stemness inherent in this cell type has now been independently verified in a number of other tumour types (recently reviewed by Clarke et al. 2006)

6 The Origins of Prostate Cancer

By treating prostate cancer as a disease of epithelium, rather than a consequence of androgen action, it is possible to reconsider its origins. The presence of an initially androgen receptor-expressing (largely luminal) phenotype in tumours has directed research towards the luminal cells as the target for oncogenic change (reinforced by the transgenic mouse studies). Whilst this hypothesis implies that the androgen-independent phenotype is the result of a forced de-differentiation from luminal to more basal characteristics, it is more logical to hypothesise that the basal cell, and indeed the stem cell within that compartment, is the target for the original oncogenic hit(s) and that the resultant phenotype is the consequence of an aberrant but close to normal differentiation pattern. There is now precedent for this in a number of other tumour systems, for example in human breast cancer where the most primitive cell appears not to express the estrogen receptor (Dontu et al. 2004), although this remains controversial (Clarke et al. 2005) and the proposed longevity of the normal stem cell would provide the life span and number of cycles of self-renewal necessary for the establishment of a founder clone within the prostate epithelium. Under a de-differentiation model, the heterogeneity of gene expression (Liu et al. 2004) and indeed genotypic changes seen in any one region of the prostate (Macintosh et al. 1998) can only be explained by a large number of independent and complex genetic changes. This could only be the result of extreme genetic instability, as a result of repeated mutagenic exposure-for example in the oral cavity where the term field cancerisation has been applied (see review by Perez Ordonez et al, 2006) - and the necessary components for various types of recombination and DNA repair-based instabilities are rather rare in prostate cancers in comparison with most other common types (Rybiki et al. 2004).

If we consider that oncogenic change in a prostate gland starts as a relatively rare event, which becomes established in a particular gland because of enhanced survival of the founder clone of stem cells (with or without the influence of stromal factors), given the presumed rarity of tissue stem cell division (sometimes considered to be quiescence), a stem cell with a proliferation/survival advantage could become established as the major population in a particular gland. A mathematical model for this has recently been published, using the better fundamental knowledge available for stem cell kinetics in human colon (Calabrese et al. 2004).

What then causes this change? There are no known mutagens in prostate, although the gland will undergo proliferative changes in response to steroids and perhaps sexual activity. For example, low androgen and high estrogen levels can result in involution and prostatic remodelling. The prostate is also profoundly sensitive to infection, as shown by the frequency of prostatitis in the human population (MacLen-nan et al. 2006). The commonest result of this is an inflammatory response, with resultant cytolysis and requirement for repair (Karin 2006). Under these conditions the 'activated' stem cell is likely to accrue an advantage, and were an element of genetic instability to be present, then further changes in response to the new environment could occur. It has been suggested that the earliest such changes are observed as prostatic inflammatory atrophy (De Marzo et al. 1998; Nelson et al. 2004). The changes could be reversible, such as chromatin remodelling and methy-lation (Yegnasubramanian et al. 2004), resulting in a flexible stem cell compartment, which responds to environmental fluctuation (Feinberg et al. 2006). Such epigenetic changes can also become more permanent. A more committed change, such as gene translocation to 'fix' an advantageous gene in an active conformation, would result in a short-term advantage, but might be deleterious or at best neutral when the status quo is restored. Such translocations are common in leukaemias, but have only recently been observed in prostate cancers, where activation of the erg oncogene is one common result in more than 50% of advanced prostate cancers (Tomlins et al. 2005). Even with a simple theoretical treatment of oncogenic initiation, the potential for hetero geneity and the 'dead-end tumours' which are frequently found in the prostates of elderly men (Franks 1954) is apparent.

7 Isolation of Prostate Cancer Stem Cells

The difficulty with this hypothesis remains the final phenotype of prostate cancers. If the mutagenesis of a normal stem cell resulting in a tumour stem cell hypothesis is correct, then one possibility is that the cancer stem cell (CSC) will retain at least some of the phenotypic properties of its nSC origin. We decided to test this hypothesis by fractionating cells from malignant human prostates, using the cell surface markers which had been so successful in sub-fractionation of normal tissues. Using originally macrodissected radical prostatectomy specimens and lymph node excisions, we obtained rare epithelial cells , which proliferated in vitro after a distinct and reproducible 'lag' phase of 10-14 days. Thereafter, on a collagen matrix containing irradiated murine stroma, under conditions which did not favour epithelial differentiation, a rapid proliferation was observed.

With this technique (Collins et al. 2005) similar cultures were derived from multiple primary prostate cancer tissue samples and several primary lymph node metastases. The properties of these cells are summarised in Table 1. Despite the invasive nature of the prostate cancer stem cells in vitro, confirmation of oncogenic origin was dependent on both in vivo xenografting and gene expression profiling of the purified cell populations. Firstly, orthotropic xenografting of fewer than 1,000 P4E6 cells with the a2p>1Hi/CD133+/CD44+ phenotype was tumour initiating when xenografted into the prostates of nude mice in a matrigel plug containing viable prostatic stroma (from a different patient) (Table 2). Significantly, the luminal (AR expressing) fraction of the same tissues (selected by expression of CD57-expressing cells) had much lower tumour-initiating capacity. This result is difficult to reconcile with the transgenic mouse models of prostate cancer, in which the TRAMP mouse was generated by expression of a powerful oncogene under the control of a luminal cell gene promoter (probasin). Other more recent models (Kasper 2005), particularly conditional knockouts of genes relevant to human disease and with no obvious reliance on AR expression,

Fig. 2. Orthotopic xenografts of P4E6 stem cells. Left panels show the mouse prostate and indicate the target for the xenografted human cell plug. Lower left panel is a haematoxylin and eosin -stained section of a P4E6 tumour-bearing prostate, indicating the morphological differences between human and mouse prostate. Four panels on the right are immunohistochemically stained (brown peroxidase stain) sections of grafted tissues as follows: A P4E6 graft showing expression of basal cytokeratin; B Section of murine prostate stained with the same anti-human basal cytokeratin antibody (negative); C P4E6 graft stained for androgen receptor (nuclear stain in luminal cells); D P4E6 graft stained for prostatic acid phosphatase (cytoplasmic stain in luminal cells) -►

are more consistent with our hypothesis. The resultant tumours derived from the primitive basal phenotype of the CSCs did express differentiated (luminal) epithelial cell proteins such as luminal cytokeratin, pro-static acid phosphatase (PAP) and AR (Fig. 2).

Secondly, an Affymetrix gene expression array analysis of RNA from two cultures derived from different cancer patients (cells amplified from an original CD133+ population) compared to three pooled benign prostatic hyperplasia (BPH) cultures revealed elevated expression of genes

Table 1 Properties of prostate cancer stem cell cultures

PCSCs have a greatly expanded life span (> 30 pd compared to a maximum of 10-12 for non-malignant tissue)

PCSCs are highly invasive (> 3-fold higher than most invasive prostate cell line PC3M)

PCSCs are highly clonogenic in vitro (2D and 3D)

PCSCs have a basal phenotype but can differentiate to an AR+ luminal phenotype

PCSC-derived cell colonies express cancer-associated genes PCSCs show some evidence of microsatellite instability In a PCSC culture the content of CD133+ cells does not vary significantly The CD133+ content of prostate tumours is not related to the stage or grade of the original tumour

previously associated with prostate cancers only in the cancer samples (Fig. 3). Furthermore, high expression of matrix metalloproteinases (not shown) provided a molecular explanation for the 2-3 fold increased invasive capacity of the stem cell-derived cultures, relative to the most invasive established prostate cancer cell line PC3M (Collins et al. 2005).

One striking feature of the long-term cultures was the constant proportion of CD133+ cells, after up to 30 population doublings and independent of the original grade or stage of the tumour. While the former proportions have been observed in other in vitro culture systems (such as neuropheres and mammospheres) for cancer stem cells, there have been reports of higher proportions of CSCs in higher-grade neural tumours (Singh et al. 2003, 2004). Thus the in vitro cultures are not optimised for cancer stem cell amplification alone. The constant CD133+ proportion is suggestive of a self-renewal mechanism whilst the bulk of the culture is an amplifying (non-stem) population as illustrated in Fig. 4. The challenge of optimising in vitro culture conditions to amplify only the stem cell population therefore remains, and a longer-term risk of cross talk between the more committed amplifying cells and the primitive basal cells is frequently observed in prostate as in keratinocyte stem cells. Indeed, induction of the terminal differentiation program can be activated by inducing culture stress and the addition of androgens (Collins et al. 2005).

Table 2 P4E6 stem cell xenografts

P4E6 cell fraction

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook

Post a comment